Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Toys

Electrical Grounding in ATX Cases? 30

Posted by Cliff from the breaking-out-of-the-beige-box dept.
FoxMulder asks: "I'm constructing a clear case made of Plexiglas for my new athlon machine. I don't know for sure how the grounding system on ATX power supplies work, but I am wondering what I can do to properly ground my machine without the use of the metal case that is so classicly accepted. I also am interested to find out which parts of the machine should be grounded, and to what. I've had a little bit of an electric bill problem when my last computer with an AT power supply wasn't grounded properly."
This discussion has been archived. No new comments can be posted.

Electrical Grounding in ATX Cases? More

Electrical Grounding in ATX Cases?

Comments Filter:
  • At home I have a full tower metal case. Fairly normal. The two sides detach.

    My problem is that, even with the metal case, I still get quite a bit of RFI and it really messes with the TV antenna. If I were to attach copper screening on the inside of the two metal sides, would this help at all? Or would it not be enough to notice?


    --Elrond, Duke of URL
    "This is the most fun I've had without being drenched in the blood of my enemies!"
  • Suggest you do like "stereo" manufacturers did in the '60s and '70s, use a regular metal case wrapped in wood veneer.
  • "It is also grounded to reduce the radiated emissions from the computer, but it would serve that purpose even if it wasn't grounded..."

    Either that or it would serve as a radiating antenna :(

    That really needs to be a strictly academic question, however, as an ungrounded metal case is a safety and fire hazard, quite possibly a violation of local building codes (they like stuff to be U.L. listed and in compliance with the National Electrical Code), and a good excuse for your homeowner's insurance company to not pay off, as well as a violation of Federal Communication Commission regulations.

    The preceeding has been a U.S.-centric electrical/electronic opinion. Adjust regulatory body names accordingly for differing geography.

  • You probably should use copper instead of aluminum. It is difficult to make good electrical connections with aluminum. With copper, you can solder the seams between the sheets of material.
  • Your stereo is broken, like most consumer electronics equipment. Shielding and filtering costs money, which most manufacturers prefer to keep for themselves. You should be complaining to the company that made your stereo.

    The typical problem is that long, unshielded speaker wires behave like an antenna, feeding the RF into the power amplifier. The RF get rectified/detected into audio and amplified by the power amplifier in the stereo. Your stereo has become a radio receiver.

  • Maybe his powersupply fried and burned down the neighborhood, and they billed it to his electric bill?
    :)


    .sig
  • If you're in an old house or one that's improperly wired, the ground on one end of a serial cable may be at a different potential than that of the other end!

    Twisted pair ethernet is designed to handle a potential difference, seeing as how in its designed implementation, it often runs from one end of an office to another, etc. But I'd trust coax ethernet less: I've been zapped a time or two, touching the BNC connector and the chassis of a computer at the same time.

    Also, something I picked up on is that ethernet shouldn't be run from one building (they might be on a different supply that's out of phase relative to the other building's supply) to another over an electrical medium as there will be a potential difference.

    Instead, you should use fibre between buildings.

  • RFI (Score:3)

    by Xenu ( 21845 ) on Tuesday August 01, 2000 @04:21PM (#886948)
    Without a metal case, you should also think about the RFI (radio frequency interference) that will be generated by the computer. You need some sort of electrically conductive coating on the plexiglas. See the previous slashdot discussion, How to Build a Clear Computer Case [slashdot.org].

  • Re:Yes and no - RFI abatement how-to. (Score:3)

    by Xenu ( 21845 ) on Wednesday August 02, 2000 @09:53AM (#886949)
    Here are some hints from looking at TEMPEST qualified (NSA approved) computer equipment:
    • Copper braid or finger stock on seams where covers are attached.
    • Lots of screws to make sure that pressure is applied to every part of a seam.
    • Connectors for external cables have built-in bypass capacitors.
    • All external cables are shielded.
    • Metal mesh over fan exhaust port.
    • No holes in case larger than, I forget, 1/4 or 1/2 wave, at the highest frequency that you are concerned with.
    • All chassis parts have good electrical connections to each other.
    • AC power input goes through a good line filter.
    • Special conductive coating used on front of CRT.
    • Metal covers over floppy drives and other drives with holes for loading media.

    The two big things are to eliminate or reduce the size of holes and to prevent RFI from getting out via the external cables and AC cord.

  • I've had a little bit of an electric bill problem when my last computer with an AT power supply wasn't grounded properly.

    Could somebody more enlightened explain how a grounding problem could cause a device to draw excessive power? Unless I'm mistaken, grounding line-powered electrical devices is just for safety. I don't see how this is possible.

  • ground the ATX PSU via the 3 wire connector, you also need to ground the mobo via the small electrically conductive holes drilled thru it..the mounting screw holes....they usually are surrounded by metal rings. attach that to a thin sheet of metal and ground that. thats usually more than enough. HDDs/CDROMs etc dont need to be grounded. ive seen ATX PSU leak current inspite of the 3 wire system..so make sure the physical metal enclosure of the PSU is grounded too.
  • Can't think of one right off hand, but then... I'm just another one of the many with a degree in Electrical and Computer Engineering (where uAmps matter).

    I haven't run into a situation where the non-grounded equipment causes problems like that (including a few computers built in/around milkcrates and such).

    --
  • With the old computer that was using so much power: did you notice any of the following:

    * Smoke
    * Extreme heat (see above)
    * Some system voltage being reported too low
    * Discoloration of the metal case....

    You can always pull components to see what's grounding, assuming its not in the power supply. Before you try to avoid grounding, make sure it is in fact the problem.


  • My problem is that, even with the metal case, I still get quite a bit of RFI and it really messes with the TV antenna. If I were to attach copper screening on the inside of the two metal sides, would this help at all? Or would it not be enough to notice?
    If you bonded the sides of the case to the ends and bottom at close intervals with something electrically conductive (like screws), you'd probably be able to reduce the leakage from the case. Unless you address the problem rising from lack of electrical contact, copper isn't going to help you.

    Unfortunately, that's not the only way for RF to leak out. Any conductor which goes through a hole in the case can carry RF right through it. To get rid of this you need to shield or bypass (or both) every cable going out of the case. Keyboard and mouse cables should have the little ferrite rat-in-boa-constrictor chokes on them right next to the case. Your video cable should be fully shielded. Bypass the cables going into your modem and sound card.
    --

  • Would it be sufficient to line the inside of the case with properly grounded aluminium foil?
    To do a really good job of blocking RFI, merely grounding the foil won't do. You need to seal all gaps in the foil so that they are connected to the adjacent foil/metal either by direct conductive connection or by such high capacitance (overlap, very tight gap) that the impedance is low enough to pass the circulating current as an almost-short.

    The point is to keep any voltages from appearing on the outside of the shield. A Faraday cage does this by surrounding the device to be shielded with a conductive barrier with any holes kept considerably smaller than the wavelength of the emissions to be blocked. A gap between two sections of foil, even a slit the height of the case, is enough to allow considerable RF emission (these are actually built deliberately and called "slot antennas"). To keep your radio and other things happy, you need to avoid anything like this. If you try it, good luck.
    --

  • Grounding/Power Issues (Score:3)

    by Matt_Bennett ( 79107 ) on Tuesday August 01, 2000 @04:45PM (#886956) Homepage Journal
    Grounding should not be a problem- Ideally, you should never use the case of a computer as a current return, it should all be done with the wires back and forth from the power supply. Most metal cases are grounded for a number of reasons- if a live AC wire were to touch it, you want the circuit breaker/fuse to blow so you don't create a safety hazard, plus if it was insulated from everything else, it could attract a static charge. It is also grounded to reduce the radiated emissions from the computer, but it would serve that purpose even if it wasn't grounded, since there is probably enough capacitance to ground at the frequencies of interest anyway. If you want shielding find some copper mesh (like window screen, but copper) and line the inside of the case with it- as long as the openings in the mesh are small, it will serve just as well as a RFI shield.

    With a truly transparent case, it would be hard, but not impossible to make a shielded case- but you can put conductors on glass- every LCD display has conductors on a transparent surface. For for a while had a (front) windshield defogger that was one solid conductor- radar detectors couldn't see through it, but you shure could.

    As to power problems? I'm guessing that this chap had his ground where the neutral ought to have been. Electric utilities frown upon drawing unequal currents from the two legs of your power, and charge a surcharge when that happens, similar to if you have a strongly reactive load- The heaters in large foundries are highly inductive, to correct the power factor, they have huge capacitor rooms to put the current and voltage back into phase. On the other hand, I can't see why a load as (relatively) small as a computer could cause this sort of surcharge.
  • ...but I am wondering what I can do to properly ground my machine without the use of the metal case that is so classicly accepted.

    It's not just classically accepted, but actually required. The metal case shields the rest of the world from the large amount of RF being emitted by various items on the motherboard. Without it, you are interfering with assorted wireless communications, and the FCC doesn't like it when you do that. Interfere with the wrong communication service, and the FCC will solve the problem to their liking, not yours.

    I've had a little bit of an electric bill problem when my last computer with an AT power supply wasn't grounded properly."

    This makes no sense to me, unless some of the CPU-generated RF noise that would normally be shunted to ground was reflecting back to one of those digital power meters. No other way I can figure the AT power supply to cause that.


    --

  • What exactly was the problem with the old AT power supply and surcharges? Feel free to message me privately, you know who I am :-o

    On another note, I think all you really need to do is run a wire to all of the motherboard mounting posts, and the outside of all drives and such, and that should be good enough. Getting a copy of the ATX specs might or might not be cheap.....

  • I have a different idea - I want to build a wooden computer case - something out of a nice darkwood that I could stain.

    Would it be sufficient to line the inside of the case with properly grounded aluminium foil? Is RF sielding pretty much the same things as the 'ground' shielding I see people doing to their electric guitars?

    -B>

  • Well, at least in theory you can stop all this RF-garbage with wrapping the whole thing in some sort of metallic fence (or something similiar). Of course this doesn't propably look very exciting in with a plexiglass case, but a friend is trying to put an old 486 to a Command&Conquer box and this is what he's propably going to do.
  • Not that I am more enlightened but what I think happened was that this power supply was somehow short-circuited and current was being drained through grounding. I've seen this happen before.
  • I must agree. Even a small hole can cause a problem. I would reccomend surrounding the case with a one inch layer of lead. Since you are considering asthetics, gold is also quite dense and conducts well. It can also be polished to a shine, and is easy to purchase. Canadian maple leaf coins melt easily into a mold with a simple propane torch. Your I/O cabling can easily be sent through optoisolators to the outside of the case (through leaded glass windows, of course). You may want to consider moving to fiber optics to reduce interference with the cheap AM radio you have on top of your computer case.
    Monitors are the source of a great ammount of RF interference, and the largest target of the TEMPEST receiver. To protect yourself and your family, place your monitor into an elegant leaded crystal monitor box. Waterford crystal would be happy to manufacture one for you.
    We all must be careful, or the feds will use a 100 million dollar surveilance van parked across the street to look at our e-mail to aunt sue. They will have to pry my chocolate chip cookie recipe from my cold dead hand, because I use PGP, SSL! My whole computer room is a sealed lead lined chamber with internal batteries for power. This way the reception on my 1972 sharp b&w TV stays clear, computer or not!

    Sorry. I am tired of all of this RFI interference BS. If your TV is messed up 'cause your 'puter has the cover of the case off, then buy it a new coat hanger. Or try using proper cabling and a decent antennae. Most HF signals on a mobo are sandwitched between groundplanes and don't radiate too much. Think of it this way, the inside of a computer case is a rats nest of little antennaes. If anything was radiating at high power, your computer wouldn't work properly. Your worries should be centered around the heat melting the plexi rather than RF. I would power down before a vacation in case something blows and causes a fire. Plexi lights up quickly, and makes a lot of noxious fumes. I had a friend set fire to a plexi window from having a candle too close. If that cheap power supply lets go, you will be in trouble! Maby metal isn't soo bad after all.

    Brian
  • Have you ever tried reporting RF problems from your neighbors? I lived for a year near a big CB operator - his CB picked up in my stereo and came out my speakers. I called about this, and got passed around and around, ultimately told that I had to _write_ a certain address in Washington DC and request a certain form....

    I just turned my stereo up louder, it did the trick.

    Fool, Uncle Sam couldn't regulate to save himself. If the truth is out there, they lost it.

  • Not that I am more enlightened but what I think happened was that this power supply was somehow short-circuited and current was being drained through grounding. I've seen this happen before.

    Yeah... In this situation, the symptom is more likely localized heating of stuff you're not expected to be getting warm, rather than a symptom at the electric bill...

    <grin> If the VGA connector to your monitor is getting hot, you've either overclocked your video card way too far, or you've got a grounding issue.

    But, seriously, make sure that your entire computer is plugged into the same outlet (circuit). That means computer, monitor, printer, sound system, etc. All of those things are connected by computer cables that have big grounds on them. If you're in an old house or one that's improperly wired, the ground on one end of a serial cable may be at a different potential than that of the other end!

    Twisted pair ethernet is designed to handle a potential difference, seeing as how in its designed implementation, it often runs from one end of an office to another, etc. But I'd trust coax ethernet less: I've been zapped a time or two, touching the BNC connector and the chassis of a computer at the same time.

    Remember, if any two grounded devices are on different circuits and the building (or extension cords, or power bars, or UPSs) are wired wrong, touching both of them at the same time could be fatal.

  • Electric utilities frown upon drawing unequal currents from the two legs of your power, and charge a surcharge when that happens, similar to if you have a strongly reactive load

    In a domestic situation, isn't the "frowing" more a feature of the way the consumption meter works?

  • On another note, I think all you really need to do is run a wire to all of the motherboard mounting posts, and the outside of all drives and such, and that should be good enough. Getting a copy of the ATX specs might or might not be cheap.....

    Sure! That'll work. Until the unmarked white van starts parking in front of your house. A few days later, a couple of nice chaps with government papers and nice trenchcoats will get out, enter your house, and remove the very nice broadband 200W or so radio transmitter than you have.

    All a computer does is turn on and off 5V (and 3.3V, and 2.8V) circuits very quickly. All of these circuits are full of stray inductance and stray capacitance (making high speed computers harder to make); the stray inductance and capacitance form wonderful resonant circuits that can spew every weird harmonic of every weird system pulsetrain to every electronic device for miles.

    Not good.

    Put your computer into a grounded metal case. Don't run your motherboard on a table. Close the computer when you're not adding or removing hardware. And that way, you won't be spewing RF noise.

    To build a clear case, you have to make a faraday cage around the computer. The easiest way is to do it using a steel case, but since you want it to be clear, I suggest that you look into lining it with grounded aluminum window screen. Notice that when you look at other clear-cased computer stuff like iMacs, the motherboard is obscured by all the sheetmetal RF shields.

    Grounding a clear computer, from an AC power safety and electrostatics standpoint is a piece of cake. But to ground it to the point where it would pass an FCC class-B RF certification is going to be a nightmare.

  • I lived for a year near a big CB operator - his CB picked up in my stereo and came out my speakers.

    Hmmm... Okay, CB works in the 27MHz band, which is for public use. While in some jurisdictions you're supposed to have a license for CB, most of the time it's not really policed.

    CB is limited to (off the top of my head) 4W TX power for narrow-band FM modulation, 12W TX power for sideband operation. Lots of people (especially truckers, who, sterotypes aside, do like to use them) hop up CB radios, allowing for a longer range.

    One way to hop it up is to stay with a 4W/12W TX setup, but to use an antenna that provides more gain. 3dB antenna gain will work in both receive and transmit, and will allow an effective radiated transmitter power of 8W/24W. 6dB antenna gain will be double that again. But, it's perfectly legal, since no more than 4W/12W of energy is being radiated, it's just being "amplified" the same way a gramophone horn works with audio.

    The other trick, illegal for sure in both the USA and Canada, is to run a linear amplifier on the CB's output stage. 4W in, 40W out. It's not uncommon to see linear amplifier stages home-built than put out over 1kW of power in the 27MHz band. When it's on a moving vehicle, it becomes pretty hard to triangulate the position of the offending transmitter.

    If your stereo was picking up noise in the tuner, it's probably because it's a poor quality tuner (not very selective), you have a poor antenna (tuned more for 27MHz than for broadcast FM) or everything is just misaligned.

    If, however, it's coming in through your stereo when you're listening to a CD or something, you're far better off ensuring that all the parts of the stereo are properly grounded through some good, thick copper wire. Make sure that every connector on the thing is clean, since a little dirt or corrosion can make a detector similar to those blued razor blade detectors in foxhole radios built by troops at war (back in WWI and WWII).

    I called about this, and got passed around and around, ultimately told that I had to _write_ a certain address in Washington DC and request a certain form....

    I'd suggest that if you had a radio license yourself and gave them your call sign, they'd have been far more interested, because by the time you have a radiotelephone license, you're usually pretty good at ferretting out a problem.

    The FCC probably didn't try too hard because of the same reasons Microsoft charges $xxx/minute for technical support: too many people call asking fairly basic questions, and they eat up a lot of labor.

    I just turned my stereo up louder, it did the trick.

    If the CB radio in the background got drowned out more by the music playing, then the problem was clearly on your end, even if the guy was running a 5kW linear amplifier into a 12dB ERP antenna. The problem was clearly occurring *after* your volume control, and therefore in the audio output stages of your stereo system.

    Cheap stereo.

    The fact of the matter is that a good stereo should run fine right in the shadow of a TV station or even an AM radio station's 50kW transmitter. You can bet your ass that the local AM radio station, in order to get any sort of coverage and longevity out of his expensive equipment, is running that transmitter at its most efficient and into a well planned and well tuned antenna system. FCC be damned, the station's general manager is going to freak out if he's always paying for new output tubes in the transmitter.

    So, buy a good stereo. This is the sort of performance difference that you can get by not buying your stereo at Wal-Mart.

    My own stereo system shies away from the cheap-assed unshielded plastic crap of today. Parts of it I built myself. No RFI issues, even though I live within a clear line of sight of both the CN Tower [cntower.ca] (which was orginally built as a transmitter tower, but also happens to be the world's tallest building, Sears and Petronas towers be damned) and a large cellular/microwave relay tower that bristles with 120MHz range dipoles.

    It consists of:

    Acoustic Research AR-4x speakers, circa 1971.

    Speaker cables, 12 gauge HPN heater cord with 1/4" connectors. (Note that this is not overpriced consumer-grade gold-plated crap like Monster Cables; mine are just solid engineering.)

    Sound A-5000 solid-state stereo amplifier, circa 1975, point to point wiring, steel chassis. (Output stages replaced with lower noise, more modern transistors.)

    Preamplifier - homebuilt, mounted into steel chassis of A-5000. Uses two 12AT7 twin-triode vacuum tubes, all parts surface-mounted on a home-etched circuit board with a ground plane. Tube filaments are powered by rectified and filtered 6V power, to reduce cathode hum. The entire preamplifier section is housed in an old steel cigar box that just happened to fit, and makes a great shield for the tubes. Why tubes? Yeah, you can do the same thing better with good MOSFETs, but when your B+ is 12V, 1.2V of induced noise is 10% of your signal. With tubes, when your B+ is 120V, the same 1.2V of induced noise is only 1% of your signal.

    NAD 3340 CD Player.

    RCA patch cables are not used to connect the CD player to the amplifier. The NAD has balanced outputs, and the amplifier has balanced inputs, so I use XLR patch.

    No tuner. I use my Sony clock-radio to listen to Howard Stern in the mornings.

    Creative Labs SoundBlaster 16 ISA - output stages were LM741, replaced them with low-noise version. Now, if only I could somehow get the D/A converter *out* of the computer...

    Audio output from the SoundBlaster is, unfortunately, unbalanced. (Besides, there are lots more things that would induce noise into a sound card, not just unbalanced patch.) Good quality (not the gold-plated Radio Shack blister-packed crap) RCA to RCA patch cable to the amplifier.

    You don't need to be that extreme to get good sound, though I will admit that MP3s have never sounded so good.

    <grin>

  • Your stereo is broken, like most consumer electronics equipment. Shielding and filtering costs money, which most manufacturers prefer to keep for themselves. You should be complaining to the company that made your stereo.

    Or, better still, throw out the Japanese-made mass-produced plastic crap, and replace it all with professional or old school steel-chassis American-made stereo equipment.

    Carver, McIntosh, ElectroVoice, Klipsche all immediately come to mind...

  • Also, something I picked up on is that ethernet shouldn't be run from one building (they might be on a different supply that's out of phase relative to the other building's supply) to another over an electrical medium as there will be a potential difference.

    Actually, it's got nothing to do with the phase of the AC power.

    In a perfect world, all transformers would provide perfect AC isolation between incoming AC power and the output of your computer's power supply. (In fact, this is tested at the factory, that's what the "HI POT" quality controls you sometimes see are for - "HIgh POTential" isolation testing.)

    Since the computers should therefore be isolated from the AC line, phase and issues like that are irrelevant. Ethernet, like everything else on your computer, is a pulse train that has been modulated onto the DC voltages that come out of your power supply.

    The big issue that you really can't trust, especially between buildings, is that they're grounded the same way. Potential differences can result there, and either blow out the varistors on your ethernet cards and hubs, or can result in nasty shocks.

    Instead, you should use fibre between buildings.

    Absolutely. Ethernet's limitation is merely caused by limitations in the consistency of the electrical power grid; it's not actually a failing of ethernet itself.

  • bad earth= highly resistive path=changes to p/s specs in relation to intended ground. though this was probably not the cause of excess current it could have been part of a more serious problem in the switchmode power supply having a couple of storage caps leaking to ground. and yes grounding is mainly for safety but it is used in other ways.

Slashdot Top Deals

"If it ain't broke, don't fix it." - Bert Lantz

Close